Skip to content
World clockEU--:--UK--:--USA--:--CN--:--PLDEFRIT中文EN

portal about AI and technologyevents · analysis · interviews · technical background

Search
LIVE
›

Labels came up with two markers: "AI" and "ai". The difference is huge

The American and international recording industry wants to label tracks on the player page, much like warnings for unsuitable content. Sony went further and taught a model to work out what percentage of a song is Beatles.

CultureAnalysisNaomi FeldmanPublished: 14 September 20266 min readSources 2
Labels came up with two markers: "AI" and "ai". The difference is huge

The Recording Industry Association of America (RIAA) and the International Federation of the Phonographic Industry (IFPI) said in July 2026 that they want to roll out a system for marking tracks made with artificial intelligence, together with Spotify and Apple Music. The proposal sets out two separate markers for two very different situations.

"AI" versus "ai"

The first marker is the word "AI" in capital letters on a black background. It goes on tracks built entirely by a model from a text prompt. It also goes on tracks where the lead vocal was synthesised by machine or key instrument parts were generated. The second is "ai", in lowercase, black on white, for songs made mostly by human hands that used AI help at specific stages. The system is meant to work like the adult content warning already in the services: the marker appears on the track page, and the artists, labels and distributors fill in the data themselves. For now the markers do not cover lyrics, compositions, music videos or cover art. The Recording Academy, the actors' and presenters' union SAG-AFTRA, the Human Artistry Campaign and the independent labels' association A2IM have also joined the project. It is a broad coalition, and a signal that the industry would rather set the standard itself before regulators do.

Sony counts percentages

In parallel, the majors are building tools to prove what went into a track. Sony Group has developed technology that can show what copyrighted music was used to create a track generated by AI. The system does not stop at a yes or no answer. It can estimate the contribution of individual bodies of work. It might state, for example, that a track uses 30 percent Beatles music and 10 percent Queen. If the model's developer agrees to cooperate, Sony pulls data from its underlying system. If the developer refuses, the technology compares the generated track with existing music and estimates the shares that way. The goal is a mechanism that would split revenue from AI music among the original authors in proportion to their contribution. The technology was built at Sony AI, the group's research unit, and its description was accepted at an international scientific conference. The same team developed a method to stop models from imitating the style of anime characters, for example from Studio Ghibli. The solution also works for video, games and characters. Sony says it has not yet decided whether to deploy it.

In Japan, music rights are split: composers, publishers and performers hold rights, and the makers of recordings separately. Sony Group controls large labels and publishers, including half of Michael Jackson's catalogue. It has not only the technology but also an interest in making the accounting work.

Comments 0

Sources

2
  1. 01美国唱片业协会拟与 Spotify、Apple Music 合作推进 AI 音乐标识 (IT之家)ZH
  2. 02索尼新技术可识别 AI 音乐中的版权内容 (IT之家)ZH

All figures and quotations in this text come from the sources listed below.

Content prepared by the editorial team with AI assistance.

Naomi Feldman

Naomi Feldman

Culture, food and lifestyle

Naomi Feldman covers culture, food, fashion and design, lifestyle and history for FLASH24, working from primary sources such as archive records, exhibition catalogues and first-person interviews rather than press releases. For food and design stories she checks prices, opening hours and production figures against at least two independent sources before filing. She spends much of her week talking to chefs, curators and designers, and marks the festival and fashion-week calendars months ahead to compare how the same event is covered elsewhere. Outside the desk she runs festival marathons and spends time in independent cinemas and film archives, which shapes how she reads cultural programming and restoration work. She does not publish a review without having seen the film, eaten the food or walked the space herself.

Newsroom →

Comments

0
  1. No comments yet — be the first.

Write a comment

Comments are public. We do not publish abuse, spam or advertising.